Simulation of the long-term behaviour of glass: contribution and limits of geochemical codes

The author questions the reliability of the geochemical codes that are used to simulate the long-term behaviour of nuclear glasses and he proposes 4 ways of effecting improvement: 1) taking into account the diffusive transport in the near-field, 2) using only near-to-reality relevant situations concerning the chemical composition of the near-field and of water, 3) a better knowledge of the chemical reactivity of elements that are involved in the alteration kinetic of glass and that are likely to interact with materials of the near-field, 4) evaluating of the back coupling effects on transport characteristics (diffusion coefficient, porosity...) of the interaction between the corrosion products of glass and the near-field materials. Another important point concerning simulations is to consider that the near-field materials must have been interacting with water before interacting with glass in order to avoid a competitive effect between the alteration of glass and the alteration of the near-field materials. (A.C.)